21.01.2014 Views

Xiao Liu PhD Thesis.pdf - Faculty of Information and Communication ...

Xiao Liu PhD Thesis.pdf - Faculty of Information and Communication ...

Xiao Liu PhD Thesis.pdf - Faculty of Information and Communication ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

fine-grained temporal constraints given the input <strong>of</strong> coarse-grained temporal<br />

constraints by service users [16, 55].<br />

For cloud workflow systems, the quality <strong>of</strong> temporal constraints can be<br />

measured by at least two basic criteria: 1) well balanced between user requirements<br />

<strong>and</strong> system performance; 2) well supported for both overall coarse-grained control<br />

<strong>and</strong> local fine-grained control. Generally speaking, there are two basic ways to<br />

assign QoS constraints, one is task-level assignment <strong>and</strong> the other is workflow-level<br />

assignment [98]. Since the whole workflow process is composed <strong>of</strong> individual tasks,<br />

an overall workflow-level constraint can be obtained by the composition <strong>of</strong> tasklevel<br />

constraints. On the contrary, task-level constraints can also be assigned by the<br />

decomposition <strong>of</strong> workflow-level constraints. However, different QoS constraints<br />

have their own characteristics <strong>and</strong> require in-depth research to h<strong>and</strong>le different<br />

scenarios. Currently, the problem <strong>of</strong> temporal constraint setting in scientific cloud<br />

workflow systems has not been well investigated.<br />

2.4 Temporal Consistency Monitoring<br />

After temporal constraints are assigned at build time, a cloud workflow instance is<br />

required to be under constant monitoring against temporal violations along the entire<br />

workflow execution at runtime. At present, temporal checkpoint selection <strong>and</strong><br />

temporal verification are the two basic steps for runtime monitoring <strong>of</strong> temporal<br />

consistency states [15]. Temporal checkpoint selection is to select specific activity<br />

points along cloud workflow execution to conduct temporal verification on the fly<br />

[18]. As for temporal verification, it is to check the current temporal consistency<br />

state <strong>of</strong> cloud workflow execution according to the pre-defined temporal consistency<br />

model [17].<br />

In recent years, many checkpoint selection strategies, from intuitive rule based to<br />

sophisticated model based, have been proposed [34, 64, 105]. The work in [34] takes<br />

every workflow activity as a checkpoint. The work in [64] selects the start activity as<br />

a checkpoint <strong>and</strong> adds a new checkpoint after each decision activity is executed. It<br />

also mentions a type <strong>of</strong> static activity point which is defined by users at build-time<br />

stage. Since temporal verification inevitably produces extra cost, a Checkpoint<br />

Selection Strategy (CSS) aims to dynamically select only necessary <strong>and</strong> sufficient<br />

18

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!